Hi
This is what I was saying to the guys here at my work. We just need a small proof that the customer isn't a spammer and we open it up. However, most of our customers are less-technical savy home folks. Did you have to prove to your ISP that you weren't spamming? If so, how did they have you do that?
There is a "passive OS fingerprinting" module for iptables (see http://ippersonality.sourceforge.net/). How about treating connections differently depending on the OS they're coming from? if(windows) then block else allow? :) Or is the OS fingerprint lost through NAT? I don't know.
Daniel